The database contains the followign files:
ET_daily_post_AKD.zip: Daily reference crop evapotranspiration (ET0) forecasts based on the ECMWF model (lead times 1-7 days) for each station after the post-processing with the affine kernel dressing (AKD) method.
ET_daily_post_AKD_45dtrain.zip: Daily ETo forecasts based on the ECMWF model (lead times 1-7 days) after the post-processing with the AKD method considering 45 days for training (instead of the nominal 30 days).
ET_daily_post_BC.zip: Daily ETo forecasts based on the ECMWF model (lead times 1-7 days) after the post-processing with the simple Bias correction (BC) method.
ET_daily_post_BMA.zip:  Daily reference crop evapotranspiration (ET0) forecasts based on the ECMWF-UKMO and ECMWF-NCEP-UKMO models (lead times 1-7 days)after the post-processing with the bayessian model averaging method (BMA) method.
ET_daily_post_NGR.zip: Daily reference crop evapotranspiration (ET0) forecasts based on the ECMWF, the ECMWF-UKMO and the ECMWF-NCEP-UKMO models (lead times 1-7 days) after the post-processing with the Non Homogeneous Gaussian Regression (NGR) (also known as EMOS) method.
ET_daily_post_NGR_45dtrain.zip: Daily ETo forecasts based on the ECMWF, the ECMWF-UKMO and the ECMWF-NCEP-UKMO models (lead times 1-7 days) after the post-processing with the NGR method considering 45 days for training (instead of the nominal 30 days).
ET_daily_raw_ECMWF.zip: Daily raw ETo forecasts based on the ECMWF (lead times 1-10 days) model.
ET_daily_raw_NCEP.zip: Daily raw ETo forecasts based on the the NCEP (lead times 1-10 days) model.
ET_daily_raw_UKMO.zip: Daily raw ETo forecasts based on the UKMO (lead times 1-7 days) model.
ET_weekly_Persistence.zip: Weekly ETo forecasts based on persistence.
ET_weekly_post_BC.zip: Weekly Eto forecasts based on the ECMWF, NCEP, UKMO (lead times 1-7 days) after the post-processing with the BC method.
ET_weekly_post_NGR.zip: Weekly Eto based on the ECMWF, NCEP, UKMO, ECMWF-UKMO and ECMWF-NCEP-UKMO models (lead times 1-7 days)forecasts after the post-processing with the NGR method.

The individual .csv files within the zip files contains:
St_num: The number of the Climate Reference Network (USCRN) weather station where the forecasts have been produced 
time_init:  Indicates the date the forecast has been produced (the measurements for a given day d were matched with the forecast values comprised between the +6 UTC hours of the day d and the +6 UTC hours of the day d+1)
time_end:	Indicates the forecasting date.
Obs: ETo "observations" (produced by considering the ground measurements of the weather variables as inputs in the Penman Montieth Eq.)
f1-fn: Memmbers of the ensemble ETo forecasts
Note: The files considering the post processing of daily ET0 forecasts with the NGR and the BMA methods include the fitting coefficients 
For details see:
Yuen, R., Baran, S., Fraley, C., Gneiting, T., Lerch, S., Scheuerer, M., and Thorarinsdottir, T.: ensembleMOS: Ensemble Model Output Statistics. R package version 0.8.2. https://CRAN.R-project.org/package=ensembleMOS, 2018.
Fraley, C., Raftery, A. E., Sloughter, J. M., Gneiting T.: EnsembleBMA: Probabilistic Forecasting using Ensembles and Bayesian Model Averaging. R package version 5.1.3. https://CRAN.R-project.org/package=ensembleBMA, 2016.

The file names with daily post-processed forecasts follow the following convention:
ET[1]_[2]_[3]_ld[4]_[5].csv (for example ETBMA_WA_Quinault_4_NE_ld1_ECMWF_UK.csv)
1 -> "raw" (to indicate raw forecasts) or "emos", "dress" or "BMA" to indicate post-processing with the NGR, AKD or BMA method, respectively.
2 -> Acronym of the station's state (for example IL for Illinois)
3 -> Name of the weather station
4 -> lead time (days)
5 -> ensemble model (EC or ECMWF -> ECMWF, GEFS -> NCEP, UK -> UKMO, EC_UK or ECMWF_UK -> ECMWF-UKMO, EC_GEFS_UK or ECMWF_GEFS_UK   -> ECMWF-NCEP-UKMO)


The file names with daily raw forecasts follow the following convention:
ET[1]_[2]_[3]_[4]_ld[4]_[5].csv (for example ETraw_MT_Lewistown_42_WSW_2014_ld7_GEFS.csv)
1 -> "raw" (to indicate raw forecasts) or "emos", "dress" or "BMA" to indicate post-processing with the NGR, AKD or BMA method, respectively.
2 -> Acronym of the station's state (for example IL for Illinois)
3 -> Name of the weather station
4 -> Year (2014, 2015 or 2016)
5 -> lead time (days)
6 -> ensemble model (EC -> ECMWF, GEFS -> NCEP, UK -> UKMO, EC_UK -> ECMWF-UKMO, EC_GEFS_UK -> ECMWF-NCEP-UKMO)

The file names with daily post-processed forecasts follow the following convention:
ET[1]_[2]_[3]_[4].csv (for example ETemos_MT_Lewistown_42_WSW_GEFS.csv)
1 -> "raw" (to indicate raw forecasts) or "emos", "dress" or "BMA" to indicate post-processing with the NGR, AKD or BMA method, respectively.
2 -> Acronym of the station's state (for example IL for Illinois)
3 -> Name of the weather station
4 -> ensemble model (EC or ECMWF -> ECMWF, GEFS -> NCEP, UK -> UKMO, EC_UK or ECMWF_UK -> ECMWF-UKMO, EC_GEFS_UK or ECMWF_GEFS_UK   -> ECMWF-NCEP-UKMO)

Glossary
ECMWF -> European Center Medium Range Weather Forecasts
NCEP  -> National Centers for Environmental Prediction Global Forecast System
UKMO ->United Kingdom Meteorological Office (UKMO)

BC -> Simple bias correction
NGR -> Non Homogeneous Gaussian Regression (also known as Ensemble Model Output Statistics EMOS)
BMA -> Bayessian Model Averaging
AKD -> Affine Kernel Dressing